A comparative study of chronic stressors effectivity on the stress response of Atlantic salmon (Salmo salar L.) post-smolt

(1)

1

(2)

2

A comparative study of chronic stressors effectivity on the stress response of Atlantic salmon

(Salmo salar L.) post-smolt

Thesis submitted for partial fulfillment of the degree Master of Aquaculture Biology

Muhammad Rahmad Royan

Faculty of Mathematics and Natural Sciences Department of Biological Sciences

University of Bergen Norway

May 2019

(3)

iii

ACKNOWELEDGEMENTS

First and foremost, I would like to humbly thank to my main supervisor, Professor Ivar Rønnestad, for accepting me to get involved in his big project, and this thesis would never be better without his strong encouragement and advice.

Also, I would like to express my sincerest gratitude to my co-supervisor, Dr.

Floriana Lai, for her 24 hour-helping hand, fruitful suggestions and discussions, and the bunches of laboratory skills she taught me. And, I wish to express my humble appreciation to Dr. Ann-Elise Olderbakk Jordal that introduced me to sophisticated instruments and basic laboratory skills.

The thesis has been part of a Research Council of Norway funded project (LeuSense). The experimental part of the thesis was conducted at Cargill Innovation Center, Dirdal, led by Dr. Anders Aksnes. I thank Drs. Marit Espe and Birgitta Norberg at Institute of Marine Research for letting me use their cortisol data from the trials.

I wish to thank to Indonesia Endowment Fund for Education (LPDP) that has supported my everyday needs during my study and life in Norway. I hope what I have got during my master’s will be beneficial for my beloved country, Indonesia.

Last but not least, I would like to convey my deepest gratefulness to my little family, especially to my dearest wife, Fitriani Juwita, for her great supports in every single moment, and my little princess, Hannah Sanaputri Royan, that always puts a big smile on me. They are my batteries. And also, for my family back home, colleagues and fellow students, and ones that I cannot mention one by one, I would like to thank to their co-operation and support.

Bergen, 6 May 2019 Muhammad Rahmad Royan

(4)

iv

ABSTRACT

Despite the fact that numerous studies in the literature have explored the effect of chronic or acute stressors on fish stress response, a comparative understanding of how different stressor types affect the Atlantic salmon post-smolt stress response is still not complete, particularly in view of potential paralog genes due to recent whole genome duplication (WGD) in salmonids. In this study we exposed Atlantic salmon post-smolt to chronic chasing, hypoxia and a combination of chasing and hypoxia for 8 days followed by an acute confinement at the end of the experiment. We investigated the stressors effectivity on expression of markers in the stress axis, considering various hypothalamic corticotropin-releasing factor (crf) and crf binding protein (crfbp) paralogs: crfssa03, crfssa14, crfssa19, crfssa29, crfbpssa01 and crfbpssa11. The results show that chronic stressors tend

to result in a more suppressed weight gain and growth rate for chronically stressed fish and reducing the magnitude of plasma cortisol levels at the end of the chronic stress exposure. In addition, we found that there is a proportional relationship between crfssa14 gene paralog and plasma cortisol level during chronic stress exposure, despite the presence of an anomaly when the novel stressor was induced. After the novel stressor was exposed, we found no proportional relationship between crfssa14 gene expression and plasma cortisol level.

We suggest that chasing can be used as an effective and logistically simple method to provoke stress in Atlantic salmon. This was the most pronounced chronic stressor shown by its vigorous effect on the higher magnitude of plasma cortisol level in chasing-exposed fish. We also suggest that crfssa14 gene paralog can be used as a marker since this was the gene where the expression was best correlated with the stress exposures used in this experiment. However, what is

(5)

v happening behind the scene of crfssa14 anomaly and how dynamic relationship between crf and crfbp needs to be investigated further.

(6)

vi

LIST OF TABLES

Table 2.1. The schedule of experiment set-up for stressing and sampling ... 12 Table 2.2. Primer sequences used in the RT-PCR. ... 18 Table 3.1. Condition factor of fish at the start (day 0) and at the end of the

experiment (day 9). ... 23 Table 4.1. The concentration and purity of RNA from various extraction protocols.

... 47

(9)

ix

LIST OF FIGURES

Figure 1.1. Simplified diagram of HSC and HPI axis in response to stressors .... 5

Figure 2.1. Illustration of fish distribution in the rearing tanks ... 11

Figure 2.2. Illustration of different types of stress exposures ... 12

Figure 2.3. Illustration of confinement stress exposure. ... 13

Figure 2.4. Illustration of blood sampling. ... 13

Figure 2.5. Salmon brain ... 14

Figure 2.6. Post-dissection hypothalamus ... 15

Figure 3.1. Weight (A) and length (B) of Atlantic salmon post smolt at the start (Day 0) and at the end of the experiment (Day 9) ... 21

Figure 3.2. Relative Growth Rate (RGR) of the Atlantic salmon post smolt. ... 22

Figure 3.3. Plasma cortisol level of Atlantic salmon post-smolt during the period of the experiment ... 24

Figure 3.4. The abundance of four crf gene paralogs in the hypothalamus of Atlantic salmon post-smolt ... 26

Figure 3.5. The abundance of two crf binding protein gene paralogs in the hypothalamus of Atlantic salmon post-smolt ... 27

Figure 4.1. Gel electrophoresis of gDNA. ... 48

Figure 4.2. Gel electrophoresis of colonies ... 49

Figure 4.3. The expression of reference genes in the experiment ... 51

(10)

1

I. INTRODUCTION

1.1. Post-smolt Salmon Production

Atlantic salmon (Salmo salar L.) has become an economically important fish commodity, and its aquaculture-related activity has been growing substantially throughout northern Europe, particularly Norway as the dominant producer (Bergheim et al., 2009). According to Norwegian Directorate of Fisheries (2018), Salmon production in Norway in recent decade had grown by 66% from 744 125 in 2007 to 1 236 353 ton in 2017. While salmon smolt is termed as newly smoltified salmon juvenile, salmon post-smolt is defined as salmon that have entered the ocean (Thorstad et al., 2012). In Norway, rearing of Atlantic salmon post-smolts to slaughtering normally takes up to 20 months in open sea cages (Aunsmo et al., 2013).

During its life cycle, especially in farming condition, Atlantic salmon might encounter different types of stress episodes. The transformation stage from parr to smolt, for instance, is known to be a typical stress-sensitive phase for Atlantic salmon since many physiological changes occur when the fish attempts to acclimatize in a higher salinity environment (Handeland et al., 1996). Later when the post-smolt have adapted to seawater, the fish will encounter numerous types of stressing conditions in aquaculture settings, such as handling, vaccination, pumping, oxygen shortage or confinement, as part of procedures for treating disease outbreak or sea lice infestation (Sveen, 2018). Not to neglect recent advancements in the technology, especially in semi-closed or closed containment culture systems, the fish also need to deal with potential stressful crowding due to intensification and high densities that are required to be economically feasible (Calabrese, 2017; Kristensen et al., 2012).

(11)

2 Despite the fact that recent advances in technology have somewhat minimized direct anthropogenic stress in fish, physical, mechanical or chemical stress will still inevitably be induced during the rearing process (Sundh et al., 2010). For example, a sea cage environment that has relatively lower current speed will not only result in slower water exchange but also cause waste accumulation that in turn suppresses the oxygen level in the sea cage area (Johansson et al., 2007; Stien et al., 2013). Moreover, when a vaccination procedure needs to be performed, the transport of fish by pumping may stimulate stress response to the fish, or when the vaccine should be administered, netting, handling and exposure to anesthetics are also unavoidable (Iwama, 1998; Kemenade et al., 2009). Quarantining the fish in a relatively small tank as part of the vaccination procedure or bath treatment may also induce confinement as well as hypoxic stressors (Gautam et al., 2017), and these simultaneous stressors can affect the biological and physiological state of the fish (Segner et al., 2012; Sundh et al., 2010). As a consequence of these prolonged stressful conditions, different whole-organism level of stress responses may appear. These include reduced growth, poor disease resistance, immune function impairment or decreased reproduction rate (Sveen, 2018). Taking together, even though better rearing-related techniques have been implemented and improved, several acute and chronic stressful conditions still exist in salmon post-smolt production.

1.2. Stress Conditions in Fish

Stress is defined as a life-threatening circumstance that can stimulate the physiological response of fish because of stressor stimuli perception (Schreck and Tort, 2016). A stressor stimulus is sometimes advantageous by enhancing performance of the fish if perceived as a mild event of stress (eustress), but it can

(12)

3 also be adaptive or maladaptive when leading to a pathological state due to its high intensity (distress) (Bonga, 1997). Since stress is inevitable in salmonids’ life, mild or less severe stress may facilitate a positive impact on fish performance by, for instance, enhancing immune system and anabolism of the fish (Dhabhar, 2008;

Sadoul and Vijayan, 2016; Yada and Tort, 2016). However, vigorous stressors often lead to impairments in numerous life aspects of the fish, such as reduced growth and appetite, increased susceptibility to disease, poor immune function or high swimming intensity (Noakes and Jones, 2016; Rodnick and Planas, 2016;

Schreck and Tort, 2016). The fact that post-smolt salmon encounter numerous types of stressor as part of husbandry activity can be a factor that elicits stress response during the production process in aquaculture system.

Based on the duration of exposure, stress stimuli can be divided into two categories: acute stressor and chronic stressor. Acute stressor is characterized as a typical short-term exposure that lasts from seconds to minutes, and the physiological response to this type of stressor depends on the severity and period of exposure (Gesto et al., 2015, 2013; Sopinka et al., 2016). For example, a study in rainbow trout and zebrafish reveals that a 3-minute chasing with a dip net resulted in 4-fold and 6 times higher plasma cortisol on stressed rainbow trout and zebrafish, respectively, relative to the control groups (Gesto et al., 2015). In a previous study, Gesto et al. (2013) found plasma cortisol of rainbow trout elevated at approximately 2, 6 and 16 times higher than that of control fish after chased for 2, 5 and 15 minutes, respectively. On the other hand, a chronic stressor is basically a prolonged exposure of a stressor during a certain period of time, it can be continuous, sequential or repeated of an acute stressor (Sopinka et al., 2016).

Chronic hypoxia (1-3 mg/l O2), for instance, was found to reduce growth of mummichog (Fundulus heteroclitus) after being exposed for 28 days relative to

(13)

4 normoxia group (7 mg/l O2) (Rees et al., 2012). In another study, the severity of chronic hypoxia is also suggested to affect channel-blue catfish weight in which the more severe the chronic hypoxia (indicated by less saturated O2), the less the weight that was found (Green et al., 2012).

A typically physical stressor, such as chasing, seems to be more pronounced in eliciting stress response compared to other type of stressors. A study in silver catfish, for instance, shows that 30-second chasing episode stimulated significantly higher plasma cortisol level compared to the exposure to agrichemical compounds, such as methyl-parathion-based insecticide, tebuconazole-based fungicide, glyphosate-based herbicide and atrazine-simazine-based herbicide (Koakoski et al., 2014). However, there are few studies as to how a physical stressor that is chronically induced is compared with other type of chronic stressor. Furthermore, despite the fact that many studies have explored the effects of a single stressor on stress response, either acute or chronic (Burt et al., 2014; Hansen et al., 2015;

Madaro et al., 2016b, 2015; Remen et al., 2014, 2012; Vikeså et al., 2017; Vindas et al., 2017b), the understanding of how simultaneous stressors affect stress response, particularly in Atlantic salmon post-smolt, is still very weak. Indeed, stressors never work alone in real aquaculture settings, instead they work in concert with other stressors. Therefore, a comparative study of how different types of stressor alone and in combination with other stressors affect the stress response of Atlantic salmon post-smolt are of importance.

1.3. Stress Response in Salmonids

There are two main stress response pathways in fish: Hypothalamic- Sympathetic-Chromaffin Cell (HSC) axis and Hypothalamic-Pituitary-Interrenal (HPI) axis (Figure 1.1). When a stress stimulus is recognized by Central Nervous

(14)

5 System (CNS), hypothalamus will be activated and triggers preganglionic sympathetic nerves that later on stimulate chromafffin cells in the head kidney to secrete catecholamines, as the incipient stress response (Bonga, 1997; Sopinka et al., 2016; Yada and Tort, 2016). HSC pathway only takes seconds until the release of catecholamines. Following the secretion of catecholamines, the production of cortisol through HPI pathway is initiated by the release of corticotropin-releasing factor (crf) hormone from the hypothalamus. This hormone will activate the formulation of pro-opiomelanocortin (POMC) in the pituitary gland which in turn will be the precursor of adrenocorticotropic hormone (ACTH) and melanophore-stimulating hormone (α-MSH). Through the blood stream, ACTH will be transported to the interrenal gland and stimulate cortisol production. Unlike catecholamines that are commonly produced within seconds, the secretion of cortisol may take from minutes to hours, thus making it more common to analyze due to the ease of method in laboratory settings (Bonga, 1997; Sopinka et al., 2016; Yada and Tort, 2016). Considering the response period, it is important to decide which pathway to choose in view of the complexity of experimental design.

Figure 1.1.Simplified diagram of HSC and HPI axis in response to stressors (Royan, 2019)

(15)

6 The stress response in fish is generally categorized into three phases: primary, secondary and tertiary stress response. As mentioned earlier, after stress stimuli are perceived the primary stress response will be induced as indicated by the upregulated catecholamines and cortisol level (Bonga, 1997; Iwama, 1998).

However, there is no fixed term as to how fish can be considered stressed or how fish can be considered in the resting state. For example, Iwama (1998) argued that plasma cortisol level below 10 ng/ml in salmonids can be considered unstressed, whilst it was found that slight and chronic upregulation of cortisol around 5-10 ng/ml was linked to the suppression of Coho salmon immune system (Maule et al., 1993).

Moreover, Nilsen et al. (2008) found a relatively high resting level of plasma cortisol (> 50 ng/ml) in Atlantic salmon after being acclimatized to a marine environment for a month. Hence, due to this unstandardized circumstance, comparisons with unstressed fish as a reference can be used to determine the stress status of fish that are exposed to stressors based on the plasma cortisol level variability.

While the primary stress response is often related to hormonal regulation, the secondary response is indicated by physiological alterations occurring in blood or tissues as a result of hormonal effects, i.e. changes in acid-base balance, blood glucose levels, immunological functions or ion balance (Bonga, 1997; Sopinka et al., 2016). For instance, Fanouraki et al. (2011) suggest that the exposure of 5-6 minutes of chasing and 1-1.5 minutes of air exposure resulted in different responses of plasma glucose level in some selected Mediterranean marine fish.

Ultimately, the tertiary stress response, also referred to as whole-organism stress response, is obvious when the fish are subjected to severe and prolonged stressors. This can be observed not only in organismal level, but also population level in which there might be effects in growth, body mass, disease resistance, reproduction or immune response of the fish (Naderi, 2018; Sveen, 2018). For

(16)

7 example, not only was the suppression of growth, weight and length found in Atlantic salmon post-smolt after being exposed to certain threshold of chronic hypoxia (Burt et al., 2014; Hansen et al., 2015; Remen et al., 2014, 2012; Vindas et al., 2017b), but similar effects are also normally observed in other fish species, such as Atlantic cod, wild Gulf killifish and channel-blue hybrid catfish (Cheek, 2011; Green et al., 2012; Methling et al., 2010; Sanchez et al., 2011). Thus, it would be interesting to see how the different stages of the stress responses are influenced by different chronic and simultaneous stress exposures.

1.4. Corticotropin-releasing Factor (CRF) and CRF-binding Protein (CRFBP) as the Regulator of Stress Response in Atlantic Salmon

Corticotropin-releasing hormone, often termed as corticotropin-releasing factor (crf), is well known for its essential role in regulating corticosteroid secretion by cascade stimulation through HPI axis pathway (Chen and Fernald, 2012;

Hauger et al., 2003). After stress stimuli recognition, crf is secreted by hypothalamus and activates POMC in the pituitary for ACTH synthesis.

Subsequently, cortisol is produced by steroidogenic cells in the interrenal gland after ACTH reaches the head kidney through blood stream (Bernier, 2006; Conde- Sieira et al., 2018; Winberg et al., 2016). The regulation of corticosteroid synthesis in HPI axis is not solely affected by crf hormone, but crf-binding protein may also have another role. Corticotropin-releasing factor binding protein (crfbp) functions to block crf from reaching pituitary gland by binding and reducing its bioavailability, thus preventing the secretion of ACTH (Geven et al., 2006; Gorissen and Flik, 2016; Huising et al., 2008; Manuel et al., 2014).

Some studies have revealed that crf mRNA expression in the preoptic area (POA) of the brain is directly proportional to the protein level of cortisol in the blood despite not always straightforward, whereas crfbp plays a role as crf blocker

(17)

8 (Sopinka et al., 2016). For instance, the elevated crf mRNA expression in Atlantic salmon post-smolt is followed by the increase in plasma cortisol compared to its resting level (Madaro et al., 2015). In addition, the upregulation of plasma cortisol after fish was being exposed to a novel stressor was confirmed by the higher abundance of crf mRNA in the POA of Atlantic salmon parr (Madaro et al., 2016b).

This phenomenon also occurs in rainbow trout in which the elevation of crf mRNA expression in cortisol-treated and subordinated fish is in line with the upregulation of plasma cortisol level (Jeffrey et al., 2012). Meanwhile, crfbp mRNA abundance was found relatively higher compared to crf mRNA expression in Atlantic salmon parr (Madaro et al., 2016b) and post-smolt (Madaro et al., 2015), albeit insignificant. Likewise, the inverse relationship between crf and crfbp mRNA expression was also observed in rainbow trout (Jeffrey et al., 2012) and Senegalese sole (Wunderink et al., 2012). These findings indicate a decrease in crf bioavailability as a result of increased crfbp peptides. Despite the fact that some

studies analyze the POA to assess the expression of crf and crfbp mRNA (Doyon et al., 2005; Ebbesson et al., 2011; Jeffrey et al., 2012; Madaro et al., 2016a, 2015;

Samaras et al., 2018), there are other primary locations of crf-related peptide expression in hypothalamus: nucleus lateralis tuberis (NLT) and nucleus recessus lateralis (NRL) (Bernier, 2006). Hence, the analysis of whole hypothalamus is required to get a comprehensive identification of crf-related peptide gene expression.

The fact that crf-related peptides are not only expressed broadly in different areas of hypothalamus but also in different parts of brain might indicate that these peptides serve different functions, despite having not been completely explored (Alderman and Bernier, 2007; Bernier, 2006; Kovacs, 2013). Interestingly, a study in spotted gar and various vertebrates, such as marsupials, monotremes, lizards,

(18)

9 turtles, birds and fishes shows that there is a duplicated homolog of crf gene (crh2) as a consequence of the second round of whole-genome duplication (WGD);

however the function of the homologs still remain unexplored (Grone and Maruska, 2015a). Due to the loss of this homolog in teleost fish during the third round of WGD, Grone and Maruska (2015b) tried to investigate another option for a possible gene duplication in teleosts, and found two paralogs of crf genes: crha and crhb.

They attempted to characterize these gene paralogs in African cichlid and zebrafish, and argued that there is probably neo-functionalization of crha paralog because of its diverse localization in different fish species.

In salmonids, as a group of teleost that have undergone the fourth round of WGD, often referred to as Ss4R (salmonids-specific 4^th vertebrate whole-genome duplication) event, a comprehensive study concerning the divergence of Ss4R gene duplicates reveals that neo-functionalization normally occurs among Ss4R duplicates (Lien et al., 2016). The Ss4R event appears to open a new chance to evolve a variety of gene duplicates with separate and important functions in stress response, particularly in Atlantic salmon post-smolt. Indeed, in our in silico analysis, we found that there are several crf and crfbp gene paralogs across the Atlantic salmon genome. The fact that many of studies that have been mentioned earlier studied only one crf and crfbp gene, creates a unique opportunity to characterize hypophysiotropic function among the gene paralogs. In other words, how these diverse gene paralogs are related to stress response in Atlantic salmon post-smolt and how they respond to different types of chronic stress exposures need to be investigated.

(19)

10 1.5. Objectives and Hypotheses

Until recently, there have been numerous studies exploring how an acute or chronic stress exposure affects the stress response in Atlantic salmon post-smolt (Anttila et al., 2013; Burt et al., 2014; Calabrese et al., 2017; Handeland et al., 1996; Hansen et al., 2015; Johansson et al., 2007; Madaro et al., 2016a, 2016b, 2015; Oldham et al., 2019; Olsen et al., 2012; Remen et al., 2014, 2012; Singer et al., 2003; Solstorm, 2017; Sveen, 2018; Vikeså, 2017; Vikeså et al., 2017; Vindas et al., 2017a, 2017b). Nonetheless, there is somewhat limited literature concerning a comparative study of different types of chronic stressors and how stressors that work in concert influence the stress response of Atlantic salmon post-smolt.

Moreover, to the best of our knowledge, there is no study so far exploring how the diversity of stress-related gene paralogs resulted from the Ss4R event is linked to the stress response of Atlantic salmon post-smolt. Therefore, this study aims to investigate how different types of chronic stressors affect the stress response of Atlantic salmon post-smolt, considering potential presence of various stress- related gene paralogs. In this study, we evaluate several response parameters, i.e.

weight, length, growth, plasma cortisol level, crf and crfbp gene paralogs, as an effect of different types of stressors.

Based on the aforementioned considerations, we hypothesize that:

H01 : Different types of stressors that are exposed have similar effects on the stress response of Atlantic salmon post-smolt.

H02 : Different gene paralogs of crf and crfbp have analogous roles in the stress response of Atlantic salmon post-smolt.

(20)

11

II. MATERIALS AND METHODS

2.1. Experimental Units and Facilities

Figure 2.1.Illustration of fish distribution in the rearing tanks (Royan, 2018).

Four weeks prior to the experiment (May 2018) Atlantic salmon (Salmo salar, L.) post-smolt of approximately 170 g were distributed into 12 tanks (volume: ca.

600 l) with 40 fish each in Cargill Innovation Center, Dirdal, Rogaland, Norway (Figure 2.1). Fish were reared at full light condition (24:0 L:D), and the tanks were supplied with flow through seawater (28 g/l) at 9^oC and oxygen saturation 90%. 2.5 dl feed (ca. 180 g; diameter 4 mm, Adapt Marine 80, Dirdal, Norway) was given four times a day (19:00-20:15, 22:00-23:15, 01:00-02:15 and 06:00-07:15) by an automatic feeder (Hølland Teknologi AS Feeder System, Florø, Norway). Salinity, temperature and oxygen saturation were monitored daily.

2.2. Experimental Design

After the acclimation period, on 11^th June 2018 (Day 0) tanks were randomly labelled according to one of the four treatments (3 replicates/treatment): control (C), chasing (SA), hypoxia (SB) and the combination of chasing and hypoxia (SC).

Stressors were induced twice per day at around 8 am in the morning and around 3 pm in the afternoon for 9 days (day 0 – day 8). On day 9, all groups, including

(21)

12 control, were exposed to a novel stressor in the morning. Due to the complexity of the experiment set-up and the logistics involved with sampling, the protocol was applied from 11^th to 20^th June for group SA and SB and from 13^th to 22^nd June for group C and SC (Table 2.1).

Table 2.1. The schedule of experiment set-up for stressing and sampling

Date 11/6 12/6 13/6 14/6 15/6 16/6 17/6 18/6 19/6 20/6 21/6 22/6 Stressing SA

SB SA SB

SA SB SC

SC SC

Day- Day0 Day1 Day0 Day1 Day8 Day9 Day8 Day9

Sampling SA SB

SA SB

C SC

- - - - SA

SB SA SB

C SC

C SC Details: SA = Stressor A (Chasing); SB = Stressor B (Hypoxia); SC = Stressor C (Chasing + Hypoxia);

C = control

Figure 2.2. Illustration of different types of stress exposures (Royan, 2018).

As illustrated in Figure 2.2, fish belonging to group SA were chased with a brush stick for 5 minutes. Hypoxia was applied to group SB by completely closing the water inflow and reducing 2/3 of water in the tank. Once the oxygen saturation reached 55%, 5 minutes were recorded before opening the water inflow again. SC group was treated by combining chasing and hypoxia at the same time. As soon as the oxygen saturation reached 55%, the 5-minute countdown along with the chasing started. On day 9, confinement was performed as a novel stressor by

(22)

13 transferring the fish into a small bucket (40 x 20 x 20 cm) with 12 l water for 15 minutes (Figure 2.3), and then the fish were collected after 45 minutes. Oxygen saturation was recorded by using OxyGuard® Dissolved Oxygen Probe (OxyGuard International A/S, Farum, Denmark).

Figure 2.3. Illustration of confinement stress exposure (Royan, 2018).

2.3. Sampling Procedure

Sampling was carried out on day 0, 1, 8 and 9. Two and five fish per tank were sampled before and after stressors respectively on day 0 while five fish were sampled on day 1, 8 and 9 (Table 2.1). Fish were anesthetized with 300 mg/l of Tricaine Pharmaq (PHARMAQ Ltd., Hampshire, United Kingdom) in 12 l of seawater, and blood was collected immediately before length and weight were recorded.

Figure 2.4. Illustration of blood sampling (Royan, 2018).

(23)

14 The blood collection was performed by a caudal venous puncture using a vacuum syringe and BD Vacutainer® set (Ref. 367614, Becton Dickinson, Plymouth, United Kingdom). The blood was then stored overnight at 4 ôC before being centrifuged at 1000 g for 10 minutes (4 ôC) using Hettich Zentrifugen Universal 320R (Hettich®, Tuttlingen, Germany). The supernatant (serum) was collected and stored at 80 ôC until further analysis (Figure 2.4). Brain and pituitary were collected (see Appendix A) and stored in separated tubes containing RNA later (1.3 ml for Brain; 700 µl for pituitary). Samples were then stored at 4 ôC overnight prior to being transferred to -80 ôC for long-term storage.

2.4. Brain Dissection

Prior to gene expression analysis, brain samples were dissected into 9 parts:

olfactory tract, olfactory bulb, telenchepalon, pineal gland, optic lobe, cerebellum, medulla oblongata, saccus vasculosus, hypothalamus and optic nerve (Figure 2.5). Considering the primary source of crf-related genes in hypothalamus, we decided to study the whole hypothalamus and dissected it referring to the brain dissection procedure in Appendix B.

Figure 2.5. Salmon brain. A: schematic drawing; B: real image (Royan, 2018).

(24)

15 To ensure high RNA yield and good tissue integrity, the brain was placed on ice block during dissection. The brain was cleaned from membranes and blood vessels using forceps before cutting a particular part of the brain. Saccus vasculosus was the first part that could be collected easily by forceps. Pineal gland was directly removed using forceps while olfactory bulb and tract were cut using scalpel to separate it from telenchepalon. The next parts that was collected was telenchepalon and cerebellum, respectively. Prior to cutting the hypothalamus, medulla oblongata was removed, and the hypothalamus was separated away from the optic nerve before cutting. After the dissection, the hypothalamus looked like the following figure:

Figure 2.6. Post-dissection hypothalamus. A: schematic drawing;

B: real image (Royan, 2018).

2.5. Growth Rate and Condition Factor (K) Calculation

Due to its reliability in comparison to other methods and its suitability for this study, the Relative Growth Rate (RGR) was applied to calculate fish growth rate.

RGR is basically the percentage of body mass gain during certain period of time (Lugert et al., 2016). Initial weight from 2 and 5 sampled fish from each tank on day 0 was measured in addition to the final weight from 5 sampled fish from each

(25)

16 tank on day 9. Based on the initial and final weight, RGR was calculated using Equation 1.

𝑅𝐺𝑅 = ^𝑤^𝑡^−𝑤^𝑖

𝑤𝑖 × 100 (1),

in which wt is the final weight while wi is the initial weight (Lugert et al., 2016).

Meanwhile, to demonstrate the fitness of the fish after stress exposures, condition factor (K) was used and quantified using weight and length of the fish by the following equation:

100

^𝑊

𝐿³ (2),

where W is the weight (g) and L is the length of the fish (cm) (Froese, 2006).

2.6. Plasma Cortisol Analysis

Plasma cortisol measurement was done by Drs. Marit Espe and Birgitta Norberg at Institute of Marine Research using Enzyme Linked Immunosorbent Assay (ELISA) with Ellman’s reagent (see Sokolowska et al., 2013).

2.7. RNA Extraction

Three out of five sampled fish on day 0 before and after stress exposure, 1, 8 and 9 were randomly selected for gene expression analysis. The RNA extraction was done using RNeasy® Mini Kit protocol with On-column DNase Digestion (QIAGEN, Hilden, Germany). The hypothalamus was firstly put into a 2 ml tube containing 600 µl Buffer RLT and 6 µl β-Mercaptoethanol in addition to 0.6-0.7 g of zirconium oxide beads (Bertin Technologies, Versailles, France; diameter 1.4 µm) and then homogenized using Precellys 24 Homogenizer (Bertin Technologies, Versailles, France) for 15 seconds at 5,000 rpm. The other components, such as 70% ethanol, 700 µl Buffer RW1 and 1 ml Buffer RPE, were used in later steps

(26)

17 according to the manufacturer’s instruction. Afterwards, the concentration and purity of RNA were checked using NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Massachusetts, USA).

2.8. cDNA Synthesis

To avoid genomic DNA remnants, TURBO DNase-free Kit® (Ambion Applied Biosystem, Foster City, CA, USA) was used as a treatment for 1.5 µg of RNA sample before performing cDNA synthesis. Afterwards, cDNA synthesis was carried out using SuperScript^TM III Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA) in which the following components were added in a total reaction volume of 20 µl: 1 µl Oligo(dT)20 (50 µM), 1 µl of 10 mM dNTP Mix (10 mM each dATP, dGTP, dCTP and dTTP at neutral pH), 10 pg - 5 µg of total RNA (volume depends on RNA concentration), distilled water and SuperScript^TM III RT Master Mix (4 µl of 5x First-Strand Buffer, 1 µl of 0.1 M DTT, 1 µl of RNaseOUT^TM Recombinant RNase Inhibitor and 1 µl of SuperScript^TM III RT).

2.9. RT-PCR Primer Design

Primers used for Real-Time Polymerase Chain Reaction (RT-PCR) assays in this study, i.e. crfssa03, crfssa14, crfssa19, crfssa29, crfbpssa01 and crfbpssa11, were designed by Lai, F. (unpublished sequence) while ef1α (Valen et al., 2011) and SsS20 (Olsvik et al., 2005) were used as reference genes. A total of four and two gene-specific RT-PCR primer pairs were designed for crf and crfbp from Atlantic salmon sequences retrieved from the NCBI data base (https://www.ncbi.nlm.nih.gov/, see Table 2.3 for accession number). For each amplicon, primers were designed using Primer3 (http://primer3.ut.ee/) and NCBI primer designing tool (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) and

(27)

18 synthesized by Sigma-Aldrich (Saint-Louis, Missouri, USA). In addition, to avoid amplification of genomic DNA, the primers were designed to span between exon- exon junction. The primers had been analyzed for crossing point (Cq), primers efficiency (E) and melting peaks, and the products were run on a gel electrophoresis and sequenced at the sequencing facility at the University of Bergen. Primers had a single melting peak indicating good specificity and good efficiency based on the result shown by the standard curve in RT-PCR test (Appendix C). Furthermore, gel electrophoresis test also confirmed that the primers amplify amplicons with corresponding sizes as shown in Table 2.2.

Table 2.2. Primer sequences used in the RT-PCR.

Gene Primer Sequence (5’  3’) Amplicon (bp)

Accession

Number Reference ef1α F: GAGAACCATTGAGAAGTTCGAGAAG 71 AF321836 Valen et al.

(2011) R: GCACCCAGGCATACTTGAAAG

SsS20 F: GCAGACCTTATCCGTGGAGCTA 85 NM_001140843.1 Olsvik et al.

(2005) R: TGGTGATGCGCAGAGTCTTG

crf ssa03

F: GCACTTGATCCATTCCACAA 232 NM_001141590.1 XM_014190344.1

Lai, F., unpublished sequence R: ACCGATTGCTGTTACCGACT

crf ssa14

F: TGGACATATTCGGGAAATGAA 229 XM_014139989.1 XM_014139988.1

Lai, F., unpublished sequence R: GTCAACGGGCTATGTTTGCT

crf ssa19

F: AACACTTGTCGCGGGTCTTG 174 XM_014159556.1 Lai, F., unpublished sequence R: GTCGGGATCAACAGGAATCTTCA

crf ssa29

F: TCCATCACTCGTGGAAAAGGA 91 XM_014181363.1 Lai, F., unpublished sequence R: CAGGGGTTCAACGAGATCTTCA

crfbp ssa01

F: AATGGCCCCGCCCAGAT 197 NM_001173799.1 Lai, F., unpublished sequence R: ATATAGGAGGTGGAGAGATAGAT

crfbp ssa11

F: AACGGTCCCGCCCAGAT 194 XM_014128333.1 Lai, F., unpublished sequence R: TAGGTGGCAGATAGATAAAG

2.10. Real Time - PCR (RT-PCR)

Each of 20 µl RT-PCR reaction consisted of 10 µl of SYBR Green I Master Mix (Roche Diagnostic, Basel, Switzerland), 0.6 µl forward and reverse primers each (10 mM), 6.8 Ultra-Pure Water (Biochrom, Berlin, Germany) and 2 µl cDNA template. The reaction mixes were run in duplicates and loaded into 96-well plate

(28)

19 (Bio-Rad Laboratories, CA, USA), including non-template control (NTC), no- reverse transcriptase control (NRT) and positive control. The following RT-PCR protocol was performed: 1) 95 ôC for 30 seconds, 2) 95 ôC for 5 seconds, 3) 60 ôC for 25 seconds, 4) repeating step 2-3 for 39 more times. The RT-PCR was performed using C1000 Touch Thermal Cycler, CFX96 Real-Time System (Bio- Rad Laboratories, CA, USA) in connection to CFX Manager Software version 3.1 (Bio-Rad, Laboratories, CA, USA). Since the expression of both reference genes, i.e. ef1α and SsS20, was assumed to be not stable (APPENDIX D), the expression of each target gene, i.e. crfssa03, crfssa14, crfssa19, crfssa29, crfbpssa01 and crfbpssa11, represents the copy number of the corresponding target gene.

2.11. Statistical Analysis

Statistical analyses were performed using R Software System version 3.50 (The R Foundation for Statistical Computing, Vienna, Austria). All datasets were tested for the normality using Anderson-Darling Normality Test while Levene’s Test was performed to test the homogeneity of variance. In case of very significant normality and/or variance, any unprecedented outliers were removed, and the dataset were subsequently square-rooted transformed before performing the comparison test. The level of significance was set to 0.05. The effect of stressor on RGR was evaluated using One-Way ANOVA test. Meanwhile, the interaction of stressor and observation period in weight, length, plasma cortisol level and gene expression were assessed using Two-Way ANOVA test. Pair-wise multiple comparison test with Bonferroni correction was used to see differences in weight and length. Whereas, multiple comparisons test in RGR, plasma cortisol and gene expression were assessed using Tukey HSD post hoc test. All data in tables and

(29)

20 figures are provided as mean ± SEM (Standard Error of Mean) unless otherwise stated.

(30)

21

III. RESULT

3.1. Effect of Stressors on Weight, Length and Growth Rate

Figure 3.1. Weight (A) and length (B) of Atlantic salmon post smolt at the start (Day 0) and at the end of the experiment (Day 9). Bars represent means ± S.E.M; Number of fish: N = 21 on day 0 and N = 15 on day 9. Asterisk indicates the degree of significance (Two-way ANOVA followed by pair-wise multiple comparison test with Bonferroni correction; ** p < 0.01).

(A)

(B)

(31)

22 Figure 3.2. Relative Growth Rate (RGR) of the Atlantic salmon post smolt.

Relative Growth Rate (RGR) is defined as a percentage of body mass gain during certain period of time. Values represent means ± S.E.M (N = 21 for initial weight, N = 15 for final weight).

After 9 days, control fish tended to have a higher increase in body mass in contrast to the chronically stressed fish. Fish in the control group grew from 263.38

± 9.25 g on day 0 to 309.73 ± 11.33 g on day 9. Fish treated with the chasing stressor had grown by 3.10 ± 1.89 g at the end of experiment while fish exposed to hypoxia and the combination of chasing and hypoxia gained 8.82 ± 2.65 g and 3.65 ± 2.38 g, respectively (F1,138 = 3.2738, p(day) = 0.0725; Figure 3.1). These results are in line with the RGR of the fish, albeit insignificant. Control fish grew around 17.5 ± 3.5 % during the experiment while fish in the chasing, hypoxia and the combination of chasing and hypoxia had grown by around 1.19 ± 1.79 %, 3.95

± 7.09 %, 1.45 ± 3.1 %, respectively (Figure 3.2). With respect to the length, control fish grew from 30.01 ± 0.35 cm on day 0 to 31.43 ± 0.37 cm at the end of experiment; Fish belonging to chasing, hypoxia and the combination of chasing and hypoxia group had grown by 2.43 ± 0.11 cm, 2.42 ± 0.08 cm and 0.39 ± 0.07

(32)

23 cm, respectively (F1,138 = 29.3242, p(day) < 0.0001; Figure 3.1). There is no tank effect on either weight or length. While chronically stressed fish have reduced condition factor, the control fish shows a slight increase. For 10 days, there is a significant interaction effect of treatment and the observation period on condition factor (K) of the fish (F3,135 = 53.475, p < 0.0001). Control fish tended to exhibit an elevation in condition factor from 0.967 ± 0.008 on day 0 to 0.992 ± 0.009 on day 9, albeit insignificant. On the other hand, chronically stressed fish show a significant decline in condition factor for chasing as well as hypoxia group, whereas the combination of chasing and hypoxia group tended to show a reduction in condition factor despite insignificant (Table 3.1).

Table 3.1. Condition factor of fish at the start (day 0) and at the end of the experiment (day 9). Condition factor (K) is defined as the fatness of the fish considering its body weight and fork length. Values represent mean ± S.E.M.

Asterisk indicates the degree of significance (Two-way ANOVA followed by Tukey’s post hoc test; ns p > 0.05, **** p < 0.0001).

Treatment Period Condition Factor N Significance Degree Control Day 0 0.967 ± 0.008 21

Day 9 0.992 ± 0.009 15 ns

Chasing Day 0 1.145 ± 0.012 21

****

Day 9 0.911 ± 0.011 15

Hypoxia Day 0 1.134 ± 0.013 20

****

Day 9 0.925 ± 0.014 15

Chasing + Hypoxia

Day 0 0.974 ± 0.016 21

Day 9 0.958 ± 0.013 15 ns

(33)

24 3.2. Plasma Cortisol

Figure 3.3. Plasma cortisol level of Atlantic salmon post-smolt during the period of the experiment. Fish were exposed to three different types of chronic stressors from Day 0 Post-Stress to Day 8, and on Day 9 all groups were exposed to a novel stressor (confinement). Bars represent mean ± S.E.M. (N = 6 for each group on day 0 before stress; N = 15 for each group on the rest of observation period). A Two Way ANOVA test shows a significant interaction effect (stressors x day of the experiment): F12,226 = 12.938, p < 0.0001. Asterisks represent the significance degree quantified by Tukey’s post hoc test (* p < 0.05, ** p < 0.01, ***

p < 0.001, **** p < 0.0001).

For all groups, plasma cortisol elevation triggered by chronic stressors on day 0 is lower on day 8 and surged on day 9 after a novel stressor. The plasma cortisol begins with no significant difference among groups on day 0 before stressor exposure. Chronic stresses initiated on day 0 appear to elevate the plasma cortisol of stressed groups 1 hour after, but the control group that was left unstressed also shows a rise. Nevertheless, unlike the plasma cortisol of stressed groups that still remains elevated, that of the control group plunges to the basal level after 24 hours.

There are significant differences between groups in this period as shown by the comparisons of each stressed group toward control group (F12,226 = 12.938, p <

(34)

25 0.0001; Figure 3.3). The magnitude of plasma cortisol of the stressed groups decreases on day 8 despite having been continuously exposed to stressors for a week. Meanwhile, the extreme upsurge of plasma cortisol in all groups including control is observed after the novel stressor exposure, in which the control group leads as the highest (142.7 ± 8.31 ng/ml) followed by hypoxia group (134.6 ± 6.45 ng/ml), chasing (132.85 ± 9.46 ng/ml) and the combination of chasing and hypoxia (123.44 ± 10.22 ng/ml). In addition to be significantly different with respect to the interaction effect (treatment x observation period) (F12,226 = 12.938, p < 0.0001), the observation period also shows a significant difference in plasma cortisol level of the fish (F4,226 = 142.288, p < 0.0001). There is no tank effect on plasma cortisol level.

(35)

26 3.3. Gene Expression

Figure 3.4. The abundance of four crf gene paralogs in the hypothalamus of Atlantic salmon post-smolt. Control and three chronically stressed groups (chasing, hypoxia and the combination of chasing and hypoxia) were observed from day 0 to day 8. On day 9, all groups including control were exposed to a novel stressor. Studied gene paralogs were crfssa03 (A), crfssa14 (B), crfssa19 (C) and crfssa29 (D). Bars represent mean ± S.E.M (N = 6 for each group on day 0 before stress; N = 9 for each group on the rest of observation period), and the values derive from copy number of the gene. Asterisks show the significance degree (* p

< 0.05, ** p < 0.01, **** p < 0.0001) as analyzed by Tukey’s post hoc test.

Figure 3.4 illustrates the expression of four crf gene paralogs (crfssa03, crfssa14, crfssa19 and crfssa29) in the hypothalamus of control and three stressed

groups of fish (chasing, hypoxia and the combination of chasing and hypoxia). A significant interaction effect (treatment x observation period) was found in crfssa19 paralog (F12,147 = 1.842, p = 0.046). Significant differences in treatment (stress exposure) were observed in crfssa14 (F3,149 = 4.895, p = 0.0028) and crfssa29 (F3,152 = 4.25, p = 0.0065). In terms of observation period, only crfssa14 (F4,149 = 16.644, p < 0.0001) paralog exhibits a significant difference considering day 0 before stressor as the reference (day 0 after stress p < 0.05; after 24 hours p <

(A) (B)

(C) (D)

(36)

27 0.0001). On the other hand, the abundance of two paralogs of crf binding protein gene (crfbpssa01 and crfbpssa11) is depicted in Figure 3.5. A significant interaction effect (treatment x observation period) was shown only in crfbpssa01 paralog (F12,138 = 2.084, p = 0.0217) while there is no significant difference in either treatment (stress exposure) or observation period in crfbpssa11 paralog. There is no tank effect on all gene paralogs expression.

Figure 3.5. The abundance of two crf binding protein gene paralogs in the hypothalamus of Atlantic salmon post-smolt. Control and three chronic stressed groups (chasing, hypoxia and the combination of chasing and hypoxia) were observed from day 0 to day 8. On day 9, all groups including control were exposed to a novel stressor. Studied gene paralogs were crfbpssa01 (A) and crfbpssa11 (B). Bars represent mean ± S.E.M (N = 6 for each group on day 0 before stress; N = 9 for each group on the rest of observation period), and the values derive from copy number of the gene. Asterisks show the significance degree (* p < 0.05, ** p < 0.01) as analyzed by Tukey’s post hoc test.

(A) (B)

(37)

28

IV. DISCUSSION

4.1. Discussion of Findings

The aim of this study was to identify and characterize the stress response of Atlantic salmon post-smolt after exposure to chronic stressors for 8 days followed by a novel stressor at the end of experiment. There are three core points that can be deduced from the observations. First, the chronic stress exposures (chasing, hypoxia and the combination of chasing and hypoxia) tend to suppress the growth rate of the chronically stressed fish that resulted in lower weight gain. Second, the response of plasma cortisol level diminishes after 8 days of chronic stress exposure, whereas the introduction of novel stressor at the end of experiment stimulates a higher cortisol response in control fish in contrast to chronically stressed fish. Third, crf and crfbp gene paralogs in the hypothalamus are expressed diversely throughout the observation period. Only the crfssa14 gene seems to be linked with exposure to stressors used in this experiment.

4.1.1. Weight, Length and Growth Rate

Stress exposures, i.e. chasing, hypoxia and the combination of both, tend to result in a more suppressed weight gain and growth rate in stressed fish relative to control after 9 days. Despite insignificant, the data indicate that there is a higher growth rate and weight gain in unstressed fish compared to those that have been exposed to the long-term stressors. Earlier study has also shown that repeated chasing leads to reduced body mass of Atlantic salmon post-smolt where fish in the control group gain a significant body mass compared to chasing-exposed fish (Madaro et al., 2016b). The growth rate has also been suggested to be negatively affected by chasing stressor in salmonids and other fish species (Madaro et al., 2015; Pavlidis et al., 2015; Tsalafouta et al., 2015; Vindas et al., 2017b). Similarly,

(38)

29 there have been many studies concerning hypoxic stress toward reduced weight gain and growth rate. Burt et al. (2014), Hansen et al. (2015), Remen et al. (2014, 2012) and Vikeså et al. (2017), for instance, suggested that a hypoxic stressor (in the range from 40 to 70 % O2 saturation) can inhibit growth of Atlantic salmon post- smolt. Despite having different hypoxic threshold from Atlantic salmon, other species also show suppressed weight and growth after being exposed to hypoxia (Cheek, 2011; Green et al., 2012; Methling et al., 2010; Sanchez et al., 2011).

However, we are not aware of any previous studies that have investigated the chronic effect of simultaneous chasing and hypoxia on fish weight and growth. In fact, the study of chronic agrichemical compounds that is in concert with chasing suggests that these simultaneous stressors negatively affect fish weight (Koakoski et al., 2014). Therefore, we suggest that simultaneous exposure of both chasing and hypoxia might be the cause of reduced weight gain and growth rate of the chronically stressed fish.

On the other hand, the trend of relatively higher weight gain and growth rate in control fish appears to be unmatched with the length. While chronically stressed fish tend to have lower weight gain and growth rate, the findings does not demonstrate this tendency with regard to length. Nevertheless, the increase in length does not necessarily indicate that there is no suppressed growth in fish, but condition factor does. Indeed, due to the fact that condition factor of the chronically stressed fish dwindles compared to that of control fish, it shows that there was a suppression of fatness in the fish after 8-day chronic stress exposure. This finding agrees with the study in Atlantic salmon post-smolt (Remen et al., 2014, 2012) showing a decrease in the condition factor of stressed fish. Likewise, a study in rainbow trout suggests that forced swimming resulted in diminished condition factor, shown by lighter weight and leaner body shape (Farrell et al., 1991). Lower

(39)

30 condition factor (K) is not only a sign of poor well-being or fitness in fish, but it is also sometimes tied with bad nutritional status despite the fact that the link is not always straightforward (Blaxter, 1988; Bolger and Connolly, 1989; Kachari et al., 2017).

Weight, length and growth as a part of stress indicators in fish is categorized into the tertiary or whole-organism response to stress (Iwama, 1998; Sopinka et al., 2016). Negative growth of fish as a result of stressors, such as chasing and hypoxia, has previously been investigated in a plethora of papers. It shows a tight connection between stress and reduced weight, length and growth. Bonga (1997) in his review asserted that in connection to stress, reduced growth results from reduced appetite and food intake, impaired food assimilation and suppressed metabolic rate. Indeed, stressors cause negative growth in fish by impairing metabolic pathways and diverting energy allocation (Iwama, 1998; Wang et al., 2009). Diminshed energy of food due to reduced appetite and food intake can cut the energy portion to growth, whereas ineffective food assimilation because of digestive system impairment leads to the increase in faecal energy resulting in decreased growth energy allocation (Wang et al., 2009). Meanwhile, the increase in O2 consumption and the reduction in heat production are typical markers for metabolic rate suppression in fish (Richards, 2009). Some of above-mentioned aspects, however, are not covered in this study due to the limitation as well as the complexity of experimental design.

4.1.2. Plasma Cortisol

The effect of stressors that were exposed to the fish on the level of plasma cortisol seem in line with the expectation based on exploration of literature data.

For instance, the data show that the chronic stress exposure leads to lower levels

(40)

31 of plasma cortisol on day 8 as well as the sudden increase in this corticosteroid after being introduced with a novel stressor on day 9. Started with a basal level in all groups on day 0 before the stress exposure, there was a considerable increase in plasma cortisol in all groups after the fish was exposed to stressors on day 0, including control fish that is supposed to remain at basal levels due to the absence of stressor. The resting cortisol level in the beginning of the experiment is suggested to represent a normal condition since there is no specific stress exposure in this period of time (Conde-Sieira et al., 2018; Kemenade et al., 2009).

This agrees with numerous studies finding that plasma cortisol in Atlantic salmon post-smolt stays at the resting level when no specific stressors are induced (Calabrese et al., 2017; Madaro et al., 2016b, 2015; Olsen et al., 2012; Singer et al., 2003).

The event of plasma cortisol elevation in control fish on day 0 after stress exposure appears to contradict with the theory because there was no desired stressor induced to control fish on this day. The hypothesis is that even with the sampling action by netting the salmon, the plasma cortisol will rise to some extent because this might be perceived as a stressor by the control fish. Madaro et al.

(2015) argued that sampling may also contribute to disturb HPI axis beside desired stressor that is induced to the fish. The association between brief handling/netting and upregulated plasma cortisol in zebrafish has also been reviewed by Spagnoli et al. (2016). Moreover, reviews in Barton and Iwama (1991) and Bonga (1997) clarify that sampling procedure may also contribute to plasma cortisol elevation.

On the other hand, the temporal space of sampling for control fish on day 0 pre- stress and post-stress was only approximately one hour in this experiment.

Perhaps, this might also be another reason why plasma cortisol was elevated in control fish, even without exposing the fish to a desired stressor, since upregulated

(41)

32 plasma cortisol normally happens in a relatively short time (Calabrese et al., 2017;

Faught et al., 2016). A study by Gesto et al. (2013) in rainbow trout, for example, reveals that the increase in plasma cortisol occurs in a few minutes until one hour, and it returns to resting level in a few hours. Since the sampling interval is still in the range of plasma cortisol response to stress, the upregulation of plasma cortisol in control fish was most probably due to those two factors: sampling action and its short interval.

While the plasma cortisol of chronically stressed fish remains elevated after 24 hours, that of control fish returns to basal level. The downregulation of plasma cortisol at basal level in control fish indicates that there is no stress signal perceived by the fish. Indeed, when no stressor is induced, plasma cortisol level will gradually dwindle and remain at basal level after a few hours. Studies in Coho salmon (Shrimpton and Randall, 1994) and rainbow trout (Jentoft et al., 2005; Yada et al., 2007) point out that plasma cortisol increases significantly approximately one hour after the stressor was induced and returns to basal level after 24 hours. Even plasma cortisol stays back at resting level 8 hours after a stressor was exposed to rainbow trout (Gesto et al., 2013). Furthermore, 24 hours after stress exposure regimes, the plasma cortisol level of chronically stressed fish is significantly higher compared to control fish. Based on those facts, we suggest that the effect of stressors, i.e. chasing, hypoxia and the combination of chasing and hypoxia, on plasma cortisol response is still much more pronounced than that of sampling action. Thus, the variability of plasma cortisol during this experiment is indeed mainly due to the stress treatments.

The effect of chronic stress exposure on the chronically stressed fish is apparent when observing a significant reduction of their plasma cortisol level on day 8. While control fish shows a stable basal plasma cortisol level on day 8, all

(42)

33 stressed groups exhibit a downregulation of plasma cortisol after one-week stress exposure. This downregulation commonly occurs in fish, particularly Atlantic salmon post-smolt, when exposed to chronic stressors. Studies of effects of unpredictable chronic stressors (Madaro et al., 2015) and repeated chasing (Madaro et al., 2016b) on stress response of Atlantic salmon post-smolt show a dwindling level of plasma cortisol within 5 days. Even, the declining trend also happens in other salmonid species, such as Coho salmon (Shrimpton and Randall, 1994) and rainbow trout (Kiilerich et al., 2018) after being exposed to chronic stressors. The decrease in plasma cortisol as a result of chronic stress is suggested due to habituation (Barton et al., 1987; Koolhaas et al., 2011). A study in rainbow trout and Eurasian perch showed that diminished response of plasma cortisol in chronically stressed fish caused by repeated stressor indicates habituation to the stress stimuli (Jentoft et al., 2005). However, chronic downregulation as a result of repeated stressors is sometimes interpreted to connect with impaired HPI axis reactivity due to being exhausted of mounting a proper response of cortisol (Jeffrey et al., 2014; Øverli et al., 1999). Despite having been downregulated after being exposed to chronic stress for a week, the plasma cortisol of chronically stressed fish on day 8 is still significantly higher than that of control, indicating the adverse effect of the chronic stressors on the fish.

To evaluate the effect of habituation due to chronic stress, a novel stressor was subjected to all groups, including control. Consequently, we found an upsurge of plasma cortisol level in all groups. A vigorous stress response after an acute stress exposure normally occurs in unstressed fish, but the assumption of habituation happening in the chronically stressed fish still remains vague. Madaro et al. (2016b, 2015) have clearly described the phenomenon where plasma cortisol level of chronically stressed Atlantic salmon post-smolt surges after exposure to

(43)

34 an acute stress. Besides, not only is the trend observed in Atlantic salmon parr (Madaro et al., 2016b), but it is also consistent in other salmonids as well as in other fish species, such as rainbow trout, brown trout, gilthead sea bream and European sea bass (Barton et al., 2005, 1987; Culbert and Gilmour, 2016; Jeffrey et al., 2014; Pickering et al., 1987; Samaras et al., 2018; Tsalafouta et al., 2015).

The studies mentioned above found that plasma cortisol level of control fish is higher than that of chronically stressed fish after being subjected to a novel acute stressor. Consistently, such a circumstance is also found in the current experimental data where fish that were chronically exposed to chasing, hypoxia and the combination of chasing and hypoxia seem to have more suppressed level of plasma cortisol in contrast to control fish, albeit insignificant. In other words, confinement as a novel stressor in this experiment was more pronounced to control fish compared to chronically stressed fish.

There are two arguments as to why the chronically stressed fish has lower plasma cortisol level than control fish after exposed to a novel stressor. First, as a consequence of adaptation, the physiological response of the fish tolerates the subsequent stressor through the negative feedback of HPI axis, thus resulting in reduced response to a novel stressor (Barton et al., 2005; Madaro et al., 2016b, 2015; Pickering et al., 1987). Second, the sub-level plasma cortisol is probably due to the cumulative burden of the prolonged stress that goes beyond the allostatic load of fish as exhibited by the other whole-organism stress responses: growth reduction, inhibition of reproduction and impaired immune response (Barton et al., 1987; Bonga, 1997; Haukenes and Barton, 2004). Owing to the fact that the stressor interval in current experiment is only within a few hours, our findings appear to agree with the latter argument since the cumulative stress response might occur as a result of short interval of stressor. Indeed, while the wider interval